GMEOW Spatial Aggregation Module

• IRI: https://blackcatinformatics.ca/gmeow/slices/aggregation
• Tier: extension
• Group: extensions

What This Slice Covers

This slice owns 13 terms and contributes 4 mapping or projection rows. Use it when its terms match the native fact you want to preserve; use the linkage tables to see how those facts leave GMEOW for consumer vocabularies.

Dependencies

• gmeow:slices/observations
• gmeow:slices/places

Consumers

• Spatial aggregation (count, density, k-anonymity) in the solver layer over places.

Local Map

Examples

Spatial Bins

• Source: slices/extensions/aggregation/examples/spatial-bins.ttl
• GMEOW terms: gmeow:Measurement, gmeow:Organization, gmeow:Place, gmeow:ScalarQuantity, gmeow:SpatialAggregation, gmeow:SpatialBin, gmeow:aggAverage, gmeow:aggregationFunction, gmeow:hasBin, gmeow:hasUnit
• External prefixes: xsd

# SPDX-FileCopyrightText: 2026 Blackcat Informatics® Inc. <paudley@blackcatinformatics.ca>
# SPDX-License-Identifier: CC-BY-4.0
#
# Worked example: spatial aggregation IS a measurement, with a privacy floor
# . A gmeow:SpatialAggregation is a gmeow:Measurement (so it carries the
# observation bundle: observedFeature, vantage, observationMethod, result) that
# applies a gmeow:aggregationFunction (average / count / density …) over
# gmeow:SpatialBins. gmeow:minimumPopulation is the k-anonymity floor: bins with
# fewer than that many people are NOT reported, so the aggregate can be published
# without re-identifying individuals (P12 — the safe public projection).
@prefix gmeow: <https://blackcatinformatics.ca/gmeow/> .
@prefix ex:    <https://blackcatinformatics.ca/gmeow/examples/aggregation/> .
@prefix rdfs:  <http://www.w3.org/2000/01/rdf-schema#> .
@prefix xsd:   <http://www.w3.org/2001/XMLSchema#> .

ex:censusBureau a gmeow:Organization ; gmeow:name "Regional Census Bureau"@en .
ex:metro        a gmeow:Place ; gmeow:name "Metro statistical area"@en .

# --- The aggregation: average household size over the region, by survey, with a
#     1000-person minimum-population floor for safe publication.
ex:householdAgg a gmeow:SpatialAggregation ;
    gmeow:observedFeature     ex:metro ;
    gmeow:vantage             ex:censusBureau ;
    gmeow:observationMethod   gmeow:methodSurvey ;
    gmeow:aggregationFunction gmeow:aggAverage ;
    gmeow:minimumPopulation   "1000"^^xsd:nonNegativeInteger ;
    gmeow:hasBin              ex:downtownBin ;
    gmeow:observationResult   ex:avgHouseholdSize .

# The result is a scalar with its unit (a Quantity must declare exactly one unit,
# P11); a count of persons-per-household is the dimensionless QUDT number unit.
ex:avgHouseholdSize a gmeow:ScalarQuantity ;
    gmeow:quantityValue "2.4"^^xsd:decimal ;
    gmeow:hasUnit       <http://qudt.org/vocab/unit/NUM> .

ex:downtownBin a gmeow:SpatialBin ;
    rdfs:label "Downtown census tract"@en .

Terms

Classes

Properties

Individuals

Linkages

• Rows: 4
• Projection profiles: qb
• External vocabularies: qb, rdf

Guide

Aggregation — spatial summarisation as honest measurement

Slice: https://blackcatinformatics.ca/gmeow/slices/aggregation · tier: extension Count, sum, average, density, centroid, and binning over places — every summary a claim.

A statistic is not a fact about the world; it is a measurement somebody made of a region. This slice therefore adds almost nothing — and that is its doctrine. Every aggregation is a gmeow:Measurement in the universal observation stack, so vantage, observed feature, result, confidence, granularity, and temporal scope are inherited without duplication (Principle 4: one observation spine, reused everywhere). The aggregation region is the observedFeature; the result is a gmeow:ScalarQuantity. The actual arithmetic — counting, density, centroid, binning — is performed by the solver layer (Principle 12), never materialised as asserted triples in the OWL core. The slice realises the the design Location-as-reference-frame design and the centroid shortcut cross-cutting aggregation concern.

Its Principle-15 consumer, declared in the manifest: spatial aggregation (count, density, k-anonymity) in the solver layer over places — the slice exists so the solver has typed, provenance-bearing nodes to hang its computed summaries on.

Privacy is the projection layer's job, not a parallel mechanism: gmeow:coarsenTo, gmeow:displayable false, and gmeow:hasSensitivity govern whether and at what granularity a result may be published (Principle 10 — suppress or coarsen at projection time, never delete). The k-anonymity check itself is a solver computation.

The aggregation node

gmeow:SpatialAggregation

A Measurement subkind whose observedFeature is a Place and whose observationResult is a ScalarQuantity. Because it is a measurement, a published census count and a rival survey estimate over the same region are two coexisting, vantage-bearing aggregations — no privileged figure (Principle 9). Pair with gmeow:hasReferenceFrame (the spatial frame, Principle 11) and gmeow:hasGranularity from the reused core spine.

gmeow:aggregationFunction

Functional pointer from a SpatialAggregation to the statistical function applied — exactly one function per aggregation node; a region summarised two ways is two aggregations.

gmeow:AggregationFunction

The open value vocabulary of functions (Principle 9 — individuals, never subclasses): seeds aggCount, aggSum, aggAverage, aggDensity, aggCentroid, aggMinimum, aggMaximum. Density and centroid are explicitly solver-computed; the others summarise a numeric property over the entities in the region. A new function is a fresh individual, not a schema change.

Binning

gmeow:SpatialBin

A Place subkind: a generated region that partitions space for summarisation — a grid cell, a hex bin, a census tract. Its geometry (from the places slice) defines the bin boundary; being a Place, it participates in RCC-8, reference frames, and granularity like any other region. Bins are infrastructure, not discoveries.

gmeow:hasBin

Links a SpatialAggregation to its constituent bins. Non-functional in both directions: an aggregation may span many bins, and a bin may serve many aggregations (the same hex grid reused across years of data).

The privacy gate

gmeow:minimumPopulation

The k-anonymity parameter, carried on the aggregation itself: the minimum population size (k) required for the result to be disclosed. A result failing the check is suppressed at projection time — coarsened via coarsenTo or withheld via displayable false (Principle 10) — never deleted. The threshold comparison ("publish only if count >= k") is evaluated by the solver layer (Principle 12); the OWL core records the policy, not the verdict.

Reuse, not redefinition

The slice deliberately redeclares nothing: observedFeature, observationResult, and vantage come from observations; hasReferenceFrame from places; hasGranularity, hasSensitivity, and coarsenTo from core; displayable from names. The module lists them as a documentation-only reuse block — duplication would violate Principle 4.

Dependencies

Depends on observations (the Measurement spine) and places (Place, geometry, frames). Consumed by the solver layer; no other slice depends on it — aggregation is a terminal extension by design.